My name is Austin Hillebrandt. I am 1329 alumni. I used to be in our strategy department. I had a bit of free time today, and I find this type of thing fun… so I wrote a strategy guide for this year. I’m sure it has flaws, but I hope people find it useful. The link is below. It is a google doc, feel free to comment if you find an error or have a suggestion.
Interesting paper! Good work. I do think you’re overselling the cycle times. Remembering from Kartiks presentation. The average number of robot cycles is 2.5 in most games.
Very nice job. Agreed with @mfbrechting, a 15-second average cycle, especially accounting for accuracy, is insane. It is maybe within reach of the best robots in the world this year. I don’t expect it. That’s only maybe within reach any year, and this is a “mandatory floor pickup, only 11 inconsistently bouncing balls dropped randomly center field from a high height, on an unobstructed field in which drive-by opponents can send your target flying 54ft” year.
15 seconds is 7-9 cycles depending on endgame. That’s generally world record territory. Rule-of-thumb really good robots in easier years are maybe half that. A few easy-to-find data points:
- In 2012’s world high score match by my count, 341 did 4 cycles and 254 did 3, in about 1:30.
- The floor cycler (1678) for the highest-scoring Einstein match in 2013 did 6 cycles with a last-second hang.
- The average high goal count on Einstein in 2016 (robot max capacity was 1 ball) was 16 per alliance, so maybe 5-8 per robot on Einstein.
The other thing that throws this analysis off is the assumption of 30s for hanging. Legally, you can easily take longer and are protected as soon as you contact the MID RUNG. Tactically, no MID or LOW hang should take anywhere near that long. There’s some extra time for coordinating everyone fitting in the space, but particularly 2 LOW RUNG robots shouldn’t need much of that.
The combination of the too-fast cycle time and the too-slow lower HANGS throws off the calculus for all 4 rungs. Please don’t ignore hanging. There really should not be a robot on Earth that can’t at least low hang in seconds (though of course there will be many).